Synthesis, in silico and bio-evaluation studies of new isothiocyanate derivatives with respect to COX inhibition and H₂S release profiles

Abstract

The development of H₂S-donating derivatives of non-steroidal anti-inflammatory drugs (NSAIDs) is considered important to reduce or overcome their gastrointestinal side effects. Sulforaphane, one of the most extensively studied isothiocyanates (ITCs), effectively releases H₂S at a slow rate. Thus, we rationally designed, synthesized, and characterized new ITC derivatives (I1–3 and I1a–e) inspired by the natural compound sulforaphane. The anti-inflammatory properties of these compounds were evaluated by their inhibitory activities against cyclooxygenase targets COX-1 and COX-2. Additionally, the cytotoxicity of the compounds was tested using the MTT assay on LPS-induced RAW 264.7 cells, revealing no cytotoxic effects at low doses. Notably, compounds I1 and fluorine-containing ester derivative I1c emerged as the most potent and selective COX-2 inhibitors, with selectivity indexes of 2611.5 and 2582.4, respectively. The H₂S-releasing capacities of ITC derivatives were investigated and compared with that of sulforaphane, showing that while compounds I1–3 exhibit slow and similar H₂S release to sulforaphane, the release from compounds I1a–e was not as pronounced as that of the standard. Physics-based molecular modeling studies including molecular docking and molecular dynamics (MD) simulations, binding free energy calculations and absorption, distribution, metabolism, and excretion (ADME) analyses were also conducted. MD simulations analysis underscored the crucial amino acids such as Tyr385, Trp387, Phe518, Val523, and Ser530 in the interactions between I1c hit compound and COX-2. The combined in silico and in vitro findings suggest that compounds I1 and I1c are promising NSAID candidates against selective COX-2 inhibition.

New isothiocyanate derivatives (I1–3 and I1a–e) were synthesized and screened for their anti-inflammatory activities and H₂S-releasing capacities. Compounds I1 and I1c demonstrated the most potent and selective COX-2 inhibition.

Introduction

Non-steroidal anti-inflammatory drugs (NSAIDs) are among the most commonly prescribed drugs, known for their anti-inflammatory, analgesic and antipyretic effects when administered as a single agent or as a multimodal treatment.^1,2 Despite their therapeutic efficacy and widespread use, long-term administration of these compounds is associated with several adverse effects, including renal dysfunction, platelet inhibition, and gastrointestinal mucosal damage.^3–5 Although NSAIDs have different chemical structures, they exert their anti-inflammatory, analgesic and antipyretic effects by inhibiting cyclooxygenases (COXs).⁶ COXs are responsible for the conversion of arachidonic acid into various eicosanoids, in particular prostaglandins and thromboxane.⁷ These fatty acid derivatives act locally in an autocrine and paracrine manner to modulate systemic hormones.⁸

There are two distinct isoforms of COXs in humans, COX-1 and COX-2, and their gene regulation is modulated discreetly, resulting in inducible expression of COX-2 and constitutive expression of COX-1. COX-2 is predominantly expressed at sites of inflammation, rather than the persistent expression of COX-1 particularly in the gastrointestinal (GI) tract.⁹ NSAIDs inhibit COX-1 and COX-2 to varying degrees depending on their chemical properties. However, the side effects observed in the GI tract can be attributed to the inhibition of the COX-1 isoform due to its protective activity on the GI mucosa and the increase in local irritation due to the carboxylic acid moieties of NSAIDs.¹⁰ These situations have led to the use of the lowest possible dose and/or the administration of gastroprotective agents to reduce the side effects of NSAIDs. The effectiveness of these methods is limited while still preserving several side effects.^11,12

Recently, research into the development of new NSAIDs has been based on eliminating the harmful functions of the carboxylic acid moiety, as it does not show anti-inflammatory activity and is partly responsible for GI toxicity.¹³ The main strategy for avoiding GI tract damage is COX-2-specific inhibitor development which derived rofecoxib and celecoxib, highly selective COX-2 inhibitors. Initially, it was postulated that newer COX-2 selective NSAIDs were associated with an increased risk of adverse cardiovascular events. This was due to the inability of the NSAIDs to distinguish between the inflammation-induced COX-2 and the constitutively present COX-2 found in the cardiovascular system, with the exception of the GI tract. However, it was reported that traditional NSAIDs, including naproxen and ibuprofen, can also increase the undesired cardiovascular events.^14,15

Isothiocyanates (ITCs, –N C S) are generated by the enzymatic hydrolysis of glucosinolates and mainly found in cruciferous vegetables such as broccoli, cabbage, and mustard.¹⁶ Sulforaphane (SFN) is one of the best known and most active member of the ITC family as it has many potential therapeutic properties including cardioprotective, anti-inflammatory, anticancer, and antimicrobial activities.^17–19 Synthetic and natural ITCs are also known for their hydrogen sulfide (H₂S) generation in an l-cysteine-dependent manner which have cytoprotective properties and antioxidant activities.^20–22 Due to the physiological activities and cardioprotective effects of H₂S in the cardiovascular processes, compounds that generate H₂S are considered as promising agents for drug development.²³ Despite the numerous properties of ITCs, most natural ITCs are volatile oils and some are unstable at room temperature, including sulforaphane. Therefore, they are spontaneously degraded into various inactive intermediates.²⁴

The biological importance of COX-2 inhibition in the context of NSAIDs, the above information and the pharmacological nature of SFN inspired us to synthesize eight ITC derivatives that aimed to release H₂S in aqueous environment. Our major motivation was to design and synthesize ITC derivatives with different functional groups. To achieve this aim, 2-phenylethanol derivatives (I1–3) bearing isothiocyanate (–N C S) main functional group and also –H, –Br and –CN groups were obtained at appropriate experimental conditions and investigated their potentials as COX inhibitors (Fig. 1). Upon determining the superior biological activity of compound I1 compared to the standards and compounds I2 and I3, new ester derivatives of I1 (I1a–e) containing an aliphatic or electron-donating/withdrawing aromatic moiety were designed, synthesized and characterized with spectroscopic techniques (Fig. 1). COX-1 and COX-2 inhibitory activities of the products were evaluated as the main biological target. Furthermore, the cytotoxic effects of the synthesized compounds were screened by employing MTT assay on LPS-induced RAW 264.7 cells. LPS is found on the outer surface of Gram-negative bacteria and is widely used for the activation of inflammatory cells to generate inflammatory conditions on cells.²⁵ The H₂S releasing capacities of compounds were investigated to have knowledge about mucosal cell protective effects. To gain a deeper understanding of compound–target enzyme interactions, the ITC derivatives were also examined through computational studies, including molecular docking and all-atom molecular dynamics (MD) simulations. After docking the synthesized compounds into the binding pockets of the protein targets, binding free energy calculations (i.e., molecular mechanics/generalized Born surface area (MM/GBSA)) analyses were performed following comprehensive MD simulations.

Fig. 1. The molecular structures of sulforaphane and the studied compounds (I1–3 and I1a–e).

Results and discussion

Chemistry

The synthesis of new ITC derivatives (I1–3 and I1a–e) was accomplished as depicted in Fig. 2. Compound I1 was synthesized by the reaction of 2-(4-aminophenyl)ethan-1-ol with thiophosgene (CSCl₂) in the presence of DIPEA under argon. The intermediates (2-(4-amino-3-bromophenyl)ethanol and 2-amino-5-(2-hydroxyethyl)benzonitrile) of compounds I2 and I3 were obtained in a previously reported procedure.²⁶ Briefly, the intermediate of compound I2 was obtained by a bromination reaction of 2-(4-aminophenyl)ethan-1-ol with NBS in DMF at room temperature. Also, for the intermediate of compound I3, conversion of the bromo to the cyano group was achieved using nickel(ii)cyanide in NMP for 10 min under microwave irradiation (200 °C, 200 W). Conversions of the amino group (–NH₂) to the isothiocyanate functional (–N C S) in compounds I2–3 were performed with CSCl₂ using the same procedure as that for compound I1.

Fig. 2. Synthesis of compounds I1–3 and I1a–e.

Since compound I1 showed distinctive highest activity compared to the standards and compounds I2–3, various ester derivatives (I1a–e) of this compound were also designed and synthesized. New ester derivatives (I1a–e) were obtained with mild to good yields as a result of the reaction of compound I1 and various aliphatic/aromatic acyl chlorides (isovaleryl chloride, benzoyl chloride, 4-methoxybenzoyl chloride, 4-fluorobenzoyl chloride and 4-nitrobenzoyl chloride) in the presence of pyridine in DCM solvent at room temperature.

The FT-IR, ¹H NMR, ¹³C NMR and MS spectra of all new compounds are given in the ESI.† While the absence of –NH₂ stretching and the presence of a strong N C S band at 2022–2073 cm⁻¹ were observed in compounds I1–3, these bands were found in the range of 2053–2129 cm⁻¹ in ester derivatives I1a–e. The disappearance of the –OH stretching of compound I1 at 3236 cm⁻¹ and the detection of ester C O stretching at 1733–1699 cm⁻¹ in compounds I1a–e are the major evidence of the proposed structures. In the ¹H and ¹³C NMR spectra of all the synthesized compounds, aliphatic and aromatic peaks were detected at expected regions. Consequently, spectral investigations of the synthesized compounds (I1–3 and I1a–e) were found suitable with the proposed molecular structures.

COX inhibition

COXs contribute to inflammation by metabolizing arachidonic acid into PGE₂, a major inflammatory mediator, by cascade reactions.²⁷ Increased expression levels of COXs are related to many inflammation-oriented diseases including Alzheimer's disease, Parkinson's disease, certain kinds of cancers, and cardiovascular problems.^28–31 Non-specific inhibition of COXs lead to gastrointestinal damage; however, selective inhibition of COX-2 shows relatively decreased numbers of undesired side effects. Therefore, selective COX-2 inhibitor NSAIDs are thought to be a safer and more beneficial approach for the alleviation of inflammation.³² To this extent, new ITC derivatives were screened for their ability to inhibit COX-2 enzyme activity at 10 μM. Among all the ITC derivatives, compounds I1 and I1c showed the highest enzyme inhibition activity against COX-2 with 94.69% and 93.43% at 10 μM, respectively (Fig. 3A). These findings were similar to the COX-2 enzyme inhibition activity of the reference compounds, celecoxib (CLB) and sulforaphane (SFN), 89.32% and 94.60% at 10 μM, respectively. The remaining compounds increased the enzymatic activity of COX-2 at 10 μM. This increase can be attributed to the interaction of the compounds with the enzyme in a way that promotes the catalytic activity of COX enzymes, leading to an increase in activity. The binding efficiency of the compounds to their target sites could affect their activity and increase the binding efficiency of arachidonic acid to the enzyme in the given incubation time. This may increase the enzyme activity compared to that of the control group or the compounds with inhibitory activity. Afterward, the effective compounds were evaluated for their recombinant COX-1 inhibition activity at 10 μM and COX-2 inhibition activity at 1 μM; the results are presented in Fig. 3B. According to the results, compounds I1 and I1c showed high selectivity for inhibiting COX-2 compared to COX-1, even at lower doses. The selectivity indexes (SIs) of compounds I1 and I1c are shown in Table 1, according to their COX-1/COX-2 IC₅₀ ratio as 2611.5 and 2582.4, respectively. COX-2 selectivity of both compounds was much higher than that of the known COX-2 selective inhibitor CLB (SI >403).³³ Therefore, these results indicated that compounds I1 and I1c could be used as potential lead compound candidates in the selective inhibition of COX-2 enzymes with high efficiency.

Fig. 3. (A) Effects of ITC derivatives on recombinant COX-2 enzyme at 10 μM. (B) Selectivity of COX-1/COX-2 enzymes. In (B), gray columns show COX-1 enzyme inhibition at 10 μM while black columns show COX-2 enzyme inhibition at 1 μM. All experiments were performed in triplicate. All data are expressed as mean. As positive controls, celecoxib (CLB) and sulforaphane (SFN) were examined at the same doses. C: control (no inhibitor). **p < 0.005, ***p < 0.001.

Table 1. IC₅₀ values and selectivity indexes (SIs) of I1, I1c and standards with COX enzymes.

Compounds	COX-1, IC50 (μM)	COX-2, IC50 (μM)	Selectivity indexa
I1	52.23 ± 4.61	0.020 ± 0.69	2611.5
I1c	64.56 ± 9.92	0.025 ± 1.32	2582.4
Celecoxib	24.30 ± 12.36	0.060 ± 0.85	405.0
Sulforaphane	ns	6.354 ± 2.58	—

^aSelectivity indexes (SIs) of the compounds were determined by dividing the IC₅₀ values of COX-1 and COX-2. ns: non-significant.

Assessment of cell viability

The selection of non-toxic doses of the ITC derivatives is essential for anti-inflammatory activity. ITCs have physiological properties including anti-inflammatory, anticarcinogenic and antimicrobial activities. However, it is important to study ITCs in low doses since consumption of ITCs may exhibit adverse effects such as carcinogenicity and tumor-promoting effects in animal models.^34,35 To this extent, the cell viability of murine macrophages stimulated with 1 μg mL⁻¹ LPS in the presence or absence of the derivatives was investigated at 10 and 20 μM doses. According to the results given in Fig. 4, increasing doses (0–20 μM) of the ITC derivatives decreased cell viability to under 80%, except for compounds I1b and I1e. Low doses of ITCs may increase cell survival due to their anti-inflammatory activities and ability to activate the cellular defense and repair mechanisms and induce growth factors and signaling molecules to promote cell survival and proliferation. This may explain the increase in cell viability for compounds I1b and I1e at 10 μM. High doses of ITCs induce apoptosis mainly in cancer cells; however, they can also affect healthy cells. Overall, the properties mentioned above of the ITCs may lead to increased levels of cell proliferation at lower doses and inhibition of proliferation at higher doses. The compounds did not cause cell cytotoxicity in RAW 264.7 cells at 10 μM doses.

Fig. 4. The effects of ITC derivatives on cell viability in LPS-induced RAW 264.7 macrophages at 10 and 20 μM doses. C1: control without inhibitor or LPS, C2: control with LPS. Significance is relative to the C2 control group. *p < 0.01, **p < 0.005, ***p < 0.0005.

H₂S-releasing capacities of ITC derivatives

H₂S can traverse the cell membrane, plays a role in diverse systems, and can be synthesized by mammalian tissues. As a result, it can activate specific pathways or react directly with biological targets as a gasotransmitter. It relaxes blood vessels, modulates neuronal excitability, regulates cell growth, and hyperpolarizes the cell membrane.³⁶ Steady and slow release of H₂S is beneficial for drug candidates, as faster release may cause accumulation of higher concentrations of H₂S and may increase toxicity.³⁷ The H₂S-releasing capacities of the derivatives are given in Fig. 5 along with SFN. Compounds I1, I2, I3, and SFN released 36.80, 35.46, 35.70, and 43.45 μM H₂S over a 4h period, respectively. The H₂S releases of compounds I1–3 were slow and had similar patterns to that of SFN. On the other hand, ester derivatives I1a, I1b, I1c, I1d, and I1e released 11.30, 11.16, 10.91, 11.54, and 11.65 μM H₂S over the same period, respectively, and H₂S releases of compounds I1a–e were not as high as that of the standard SFN (Fig. 5).

Fig. 5. Released H₂S concentrations of the ITC derivatives and SFN over a four-hour period.

Computational studies

The ITC derivatives were docked into the binding pocket on protein targets. Additionally, molecular docking studies of SFN with COX enzymes were performed (ESI,† Table S2 and Fig. S33). The top docking scored conformations of I1 and I1c ligands, which showed the most promising in vitro results against COX-2 target among all the synthesized compounds, are depicted in Fig. 6 and 7, respectively. Docking analysis revealed that I1 forms hydrogen bonding interactions with the Met522’s backbone of the COX-2. Other nonbonding interactions were mainly hydrophobic via Val116, Val349, Leu352, Tyr355, Leu359, Leu384, Tyr385, Phe518, Val523, Gly526, and Ala527. Compound I1c forms π–π stacking interactions with Tyr355 and a hydrogen bond with Phe518’s backbone. Other nonbonding interactions were mostly hydrophobic via Val349, Leu352, Leu359, Phe381, Ile517, Met522, Val523, and Gly526.

Fig. 6. The pose of I1 with the highest docking score is shown at the binding pocket of human COX-2. (A) and (B) depict 3D protein–ligand interactions, with (A) providing a zoomed-out view and (B) offering a zoomed-in perspective. (C) focuses on the 2D ligand interaction details at the binding pocket.

Fig. 7. The pose of I1c with the highest docking score is shown at the binding pocket of human COX-2. (A) and (B) depict 3D protein–ligand interactions, with (A) providing a zoomed-out view and (B) offering a zoomed-in perspective. (C) focuses on the 2D ligand interaction details at the binding pocket.

Celecoxib and rofecoxib were used as positive controls for COX-1 and COX-2, respectively. The docking scores of ITCs and reference compounds are detailed in Table 2. Notably, all the compounds docked with COX-2 exhibited higher predicted binding affinity compared to COX-1.

Table 2. Docking scores of ITC derivatives and reference compounds at the binding pocket of COX-1 and COX-2.

Ligands	COX-2 docking score (kcal mol−1)	COX-2 ligand effective score (kcal mol−1)	COX-1 docking score (kcal mol−1)	COX-1 ligand effective score (kcal mol−1)
I1c	−8.327	−0.397	−6.266	−0.298
I1a	−8.167	−0.408	−6.071	−0.304
I3	−7.803	−0.557	−6.529	−0.466
I2	−6.837	−0.526	−6.537	−0.503
I1d	−6.463	−0.281	−4.863	−0.211
I1	−6.433	−0.536	−6.313	−0.526
I1b	−6.426	−0.292	−4.698	−0.214
I1e	−5.015	−0.279	−4.263	−0.237
Celecoxib	−11.655	−0.448	−4.669	−0.180
Rofecoxib	−9.812	−0.446	N/Aa	N/Aa

^aCompounds for which no docking pose could be generated at the end of the docking process in the Glide program, and thus for which no docking score is available, have been denoted as N/A (not available).

Among the synthesized compounds, the top three molecules with the best docking scores against COX-2 were determined to be I1c, I1a, and I3. In addition to the docking scores, ligand efficiency scores (i.e., docking score per number of non-hydrogen atoms) were also taken into consideration. In this regard, compounds I1, I2, and I3 were found to have even better ligand efficiency scores than known COX-2 inhibitor drugs. As a result, these five compounds were selected as the most promising candidates for targeting COX-2 and serving as effective COX-2 inhibitors.

The generation of a docking pose for the interaction between rofecoxib and COX-1 was not possible due to rofecoxib not binding to COX-1. To determine the specificity of the compounds to the COX-2 enzyme, each compound was also docked to COX-1 using a methodology similar to that employed for the COX-2 protein. This binding process was carried out using the same docking procedures and conditions, allowing for a comparison of the binding properties of the compounds between COX-1 and COX-2. After obtaining the docking results, the docking scores and the ligand efficiency scores of the compounds with COX-1 and COX-2 were carefully analyzed (Table 2). In general, when ligand effective scores fall within the range of −0.3 to −0.5 kcal mol⁻¹, it may indicate that these compounds have a good binding capability and effective ligand efficiency. According to the docking results, the compounds with the lowest binding to COX-1 were identified as I1b, I1d, and I1e. These compounds had a docking score similar to that of the reference compound, celecoxib. Furthermore, considering effective ligand score values, I1b, I1d, and I1e were again identified as low scores in COX-1. Similarly, the compound I1c, which demonstrated the best docking score to COX-2, was considered to have low predicted ligand affinity in COX-1 binding as it had a ligand effective score value of −0.298 kcal mol⁻¹. Detailed interaction analysis of I1c ligand and COX-2 target protein during 100 ns MD simulations are provided in Fig. 8.

Fig. 8. COX-2 target and I1c ligand interaction analysis throughout 100 ns MD simulations. (A) Amino acid residues involved in various interactions with the ligand, such as hydrogen bonding and hydrophobic interactions are detailed along with the types of these interactions. (B) The 2D representation of the ligand is presented alongside the amino acids involved in the interactions. (C) A thorough analysis of protein–ligand contacts over a 100 ns period has been conducted and is represented as a timeline.

When evaluating the results of the MD simulations analysis, it was determined that Tyr385, Trp387, Phe518, Val523, and Ser530 are the most crucial amino acids during the 100 ns MD simulations interactions between the I1c ligand and COX-2. These findings, along with our previous study, demonstrate a shared correlation with amino acid residues acknowledged as significant by other research groups targeting COX-2 inhibition.^38–42 These residues not only prove to be effective in the central binding region but also align with the crucial amino acids involved in the interaction between COX-2 and the reference compound celecoxib. Furthermore, the observation that candidate drugs typically interact with these amino acids serves to reinforce the accuracy of our findings.

In summary, based on the evaluation of docking results and ligand efficiencies, the most effective compound, I1c has been identified for its optimal binding to COX-2 and poor binding to COX-1 according to docking results. Compounds I1a, I1, I2, and I3 have shown good binding to COX-2; however, considering the possibility of binding to COX-1, they are considered risky in terms of selectivity. In this case, detailed analyses regarding selectivity have already been substantiated in our study through in vitro experiments.

To investigate the interaction of potential hit compounds with target proteins and to understand their structure and behavior at the atomic level, 100 ns MD simulations were conducted. In the simulations, a neutralized system, the TIP3P water model, and conditions of 310 K temperature and 1.01325 bar pressure were maintained. The MD simulations play an important role in understanding how the synthesized compounds interact with the COX-1 and COX-2 targets at an atomic level. These simulations provide essential insights into the stability and dynamic behavior of these interactions over an extended period. The selection of a neutralized system, the TIP3P water model, and the specific conditions of 310 K temperature and 1.01325 bar pressure were carefully chosen to closely mimic physiological conditions. This careful choice of parameters ensures that the simulation results are not only accurate but also highly relevant to real biological environments. Such parameters are widely recognized as standard in MD simulations because they effectively mimic and simulate the behavior of biological molecules under conditions that approximate those found in living organisms.

Following the completion of MD simulations, detailed and thorough binding free energy analyses (i.e., MM/GBSA) were conducted to evaluate the predicted binding affinities of the synthesized compounds (Table 3). The average MM/GBSA scores of the references (celecoxib and rofecoxib) were considered as reference values and compared with the obtained results.

Table 3. Average MM/GBSA binding free energy values of ITC derivatives over 100 ns MD simulations.

Compound	Average COX-2 MM/GBSA score (kcal mol−1)	Compound	Average COX-1 MM/GBSA score (kcal mol−1)
I1c	−69.22	I1c	−69.77
I1b	−67.87	I1d	−67.59
I1a	−66.99	I1b	−67.00
I1d	−65.25	I1a	−63.46
I1e	−60.22	I1e	−61.18
I1	−53.41	I2	−54.69
I2	−51.06	I3	−54.68
I3	−48.82	I1	−45.56
Celecoxib	−69.85	Celecoxib	−56.93
Rofecoxib	−62.42	Rofecoxib	N/A

The consistency between the best docking scores and the MM/GBSA scores of compounds with the highest docking scores was expected in our studies and confirmed by each other. According to MM/GBSA scores, the best COX-2 binding was observed between I1c with a score of −69.22 kcal mol⁻¹. This score was very close to celecoxib's score and even better than rofecoxib's MM/GBSA score. I1b with an average MM/GBSA score of −67.87 kcal mol⁻¹, I1a with −66.99 kcal mol⁻¹, and I1d with −65.25 kcal mol⁻¹ also exhibited better predicted binding affinity than the known COX-2 inhibitor, rofecoxib. I1e and I1 showed average MM/GBSA scores of −60.22 and −53.41 kcal mol⁻¹, respectively. According to the MM/GBSA scores, ligands I1, I2, I3, I1e and I1a demonstrated binding scores against COX-1 that were comparable to that of the reference drug celecoxib.

In silico results show that the synthesized compounds exhibit similarity compared to the known inhibitor drug rofecoxib, with some even demonstrating superiority in terms of docking score, ligand efficiency, and MM/GBSA results. While in silico studies suggest the potential risk of some of these compounds inhibiting not only COX-2 but also COX-1, or the need to enhance COX-2 selectivity, it has already been established through our conducted in vitro experiments that these compounds are significantly more specific to COX-2 compared to COX-1.

In summary, according to the MM/GBSA results, compounds I1c, I1b and I1a have demonstrated stronger and more stable binding to COX-2 compared to the known inhibitor. I1c has been identified as the best candidate COX-2 inhibitor compound based on the MM/GBSA analysis. While the COX-2 binding energies of the I1 and I1e compounds are moderate, these compounds exhibit weaker binding to COX-1 and have average MM/GBSA values close to the binding energy of the reference drug. Specifically, while I1 exhibits moderate binding to COX-2, its low binding to COX-1 suggests that I1 could be a good and selective COX-2 inhibitor. Therefore, based on the in silico results, it is concluded that compounds I1c, I1b, I1a, I1, I1d and I1e could be potential candidates after considering COX-2 selectivity.

The absorption, distribution, metabolism, and excretion (ADME) properties of all compounds were predicted utilizing the QikProp software (version 3.5. New York, NY, Schrödinger LLC, 2012), which evaluates the physicochemical attributes of the ITC derivatives (ESI,† Table S1). This tool has the capability to assess the drug-likeness of the ligands in accordance with Lipinski's criteria.⁴³ All compounds comply with Lipinski's rule of five.

The investigation of therapeutic activity prediction profiles concerning cancer and inflammation was conducted using the MetaCore/MetaDrug platform (https://portal.genego.com/), with the scores demonstrated in Table 4. Values exceeding 0.5 indicate a more significant potential for ligands' therapeutic activity for inflammation and cancer.

Table 4. Therapeutic activity prediction profiles of ITC derivatives. Values in parentheses represent Tanimoto prioritization (TP) scores. TP represents the highest structural similarity (Tanimoto coefficient) between query compounds and compounds used in model development.

Ligand	Inflammation	Cancer
I1c	0.78 (55.56)	0.34 (49.03)
I1e	0.68 (45.89)	0.29 (47.30)
I1a	0.65 (61.90)	0.34 (53.15)
I1d	0.62 (58.77)	0.37 (52.32)
I1b	0.55 (53.33)	0.27 (53.33)
I1	0.45 (48.00)	0.53 (48.00)
I2	0.45 (42.15)	0.53 (42.15)
I3	0.28 (41.60)	0.38 (39.84)

Ligand I1c demonstrates the highest predicted anti-inflammatory activity with a value of 0.78, suggesting the most significant impact on inflammation. I1c was also the best candidate drug against COX-2 based on docking and MM/GBSA results. Following closely, ligands I1e, I1a, and I1d provide respective scores of 0.68, 0.65, and 0.62 for anti-inflammatory predictions. Furthermore, ligand I1b exhibits a substantial prediction of anti-inflammatory activity by surpassing 0.5. Consequently, the findings derived from binary QSAR models indicate that ester derivatives I1a–e may possess significant anti-inflammatory properties. These compounds were also the promising drug candidates against COX-2 based on docking and MM/GBSA results. Additionally, among the ligands, I1 and I2 have been identified as potentially having anticancer properties based on predictions from the anti-cancer QSAR models. Therefore, it is planned to examine the anticancer properties of these compounds in the future.

Structure–activity relationships

The structure activity relationships (SAR) of the compounds are illustrated briefly in Fig. 9. Non-substituted derivative compound I1 is the most potent and selective COX-2 inhibitor (IC₅₀ = 0.020 μM) among compounds I1–3. Besides, compounds I2 and I3 containing bromo and cyano electron-withdrawing groups did not show significant inhibition. It is determined that while compounds I1–3 have similar H₂S-releasing capacities to the sulforaphane standard, compounds I1a–e which were derived from compound I1 release lower levels of H₂S than the standard. While 4-fluorophenyl ester (I1c) and phenyl ester (I1a) derivatives have the best docking scores (−8.327 and −8.167 kcal mol⁻¹, respectively) with COX-2 enzyme, aliphatic ester derivative I1e containing isovaleryl unit has the lowest docking scores with both COX-2 and COX-1 enzymes. Fluoro-containing phenyl ester derivative I1c is the most potent and selective COX-2 inhibitor (IC₅₀ = 0.025 μM) among ester derivatives I1a–e. These good biological results for compound I1c were also confirmed by in silico docking studies and therapeutic activity prediction calculations. Furthermore, no significant COX-2 inhibition of the esters bearing methoxy (I1b) or nitro (I1d) groups on the phenyl ring were detected.

Fig. 9. Summary of the structure–activity relationships of the synthesized ITC compounds.

Conclusions

In this study, we present the design, synthesis, and characterization of eight ITC derivatives (I1–3 and I1a–e) as potential anti-inflammatory agents, focusing on their COX-2 and COX-1 inhibitory activities. Initially, compounds I1–3 were synthesized by converting the amino group to isothiocyanate group in 2-(4-aminophenyl)ethanol and its bromo and cyano derivatives. The promising anti-inflammatory activity observed with compound I1 led to the synthesis of various ester derivatives (I1a–e) through esterification reactions using compound I1 and different acyl chlorides. The compounds exhibited no cytotoxicity on LPS-induced RAW 264.7 cells at a concentration of 10 μM, although higher doses showed some cytotoxic activity. Compounds I1 and I1c demonstrated strong inhibition against COX-2 (IC₅₀ = 0.020 μM and 0.025 μM, respectively) and showed significantly better selectivity toward this enzyme (SI = 2611.5 and 2582.4, respectively) compared to the COX-2 selective reference drug, celecoxib. The H₂S-releasing capacities of compounds I1–3 were favorably slow and released concentrations were similar to that of sulforaphane. Compounds I1a–e also exhibited a slow-release profile with a markedly reduced H₂S concentration in comparison to sulforaphane. ADME studies showed that the drug-likeness of our ligands was found compatible in accordance with Lipinski's criteria. ITC derivatives exhibited promising docking and MM/GBSA results in their interaction with COX-2, showing comparable or even higher affinity and protein–ligand interaction stability compared to reference drugs, celecoxib and rofecoxib. Among these compounds, I1c, I1a, and I3 demonstrated the best docking scores against COX-2. Detailed analyses, including MD simulations and MM/GBSA scores, supported the in silico docking studies, and I1c emerged as the lead compound with the highest predicted binding affinity and protein–ligand interaction stability. Also, compound I1c emerged as having the highest score considering the therapeutic activity predictions, especially in anti-inflammatory effects. In summary, based on in silico and in vitro findings, compounds I1 and I1c have been identified as promising selective COX-2 inhibitors.

Experimental

Chemistry

General

All of the chemical reagents and solvents were purchased from Merck, Sigma-Aldrich, Alfa-Aesar and Acros for synthesis at the highest commercial quality and used as received. Electrothermal IA9100 or X-4 instruments were used to detect the melting points of products. A CEM SP Discover microwave synthesis reactor was used to apply microwave energy in the synthesis of 2-amino-5-(2-hydroxyethyl)benzonitrile. Analytical thin layer chromatography (TLC) was performed using Merck 60GF₂₅₄ plates and spots were visualized under 254 nm UV light using a CAMAG UV Cabinet. Purification with column chromatography was performed on SiliCycle silica gel as the stationary phase. A Perkin Elmer Spectrum 100 spectrometer was utilized to record infrared (FT-IR) spectra using the ATR method. Proton and carbon nuclear magnetic resonance (¹H and ¹³C NMR) spectra were recorded on Agilent 600 MHz or JEOL 400 MHz spectrometers using DMSO-d₆ as solvent. The molecular mass of the compounds was determined using Shimadzu 8040 LC/MS/MS or Thermo Orbitrap Q-Exactive HRMS instruments with ESI ionization. The purity of the newly synthesized compounds was determined by NMR analyses.

Synthesis of compounds I1–3

2-(4-Isothiocyanatophenyl)ethanol (I1)

2-(4-Aminophenyl)ethan-1-ol (0.73 mmol) was dissolved in 1 mL THF and then DIPEA (1.46 mmol, 2.0 equiv.) was added and stirred for 15 min at 0 °C under argon. Then, thiophosgene (0.88 mmol, 1.2 equiv.) in THF (1 mL) was added dropwise in 30 min at the same temperature and the reaction mixture was stirred at room temperature for 1–2 h. After the reaction completion (TLC, hexane : ethyl acetate), the solvent was removed in vacuo. Then, the yellow oily liquid was purified by column chromatography (silica gel, hexane : ethyl acetate/1 : 1) to obtain the target compound. When the resulting light-yellow viscous liquid was kept in the refrigerator at +4 °C, a cream solid was obtained with 80% yield. m.p. 44–45 °C (lit. m.p. 42–44 °C⁴⁴). IR (υ/cm⁻¹): 3236 (O–H stretching), 3073, 3049, 3037 (aromatic C–H stretching), 2918, 2872 (aliphatic C–H stretching), 2172, 2073 (N C S stretching). ¹H NMR (600 MHz, DMSO-d₆, ppm): δ 7.31 (s, 2H), 7.27 (s, 2H), 4.65 (s, broad, 1H, –OH), 3.58 (t, J = 6.1 Hz, 2H), 2.70 (t, J = 6.1 Hz, 2H). ¹³C NMR (150 MHz, DMSO-d₆, ppm): 140.5, 133.1, 130.8, 128.0, 126.0, 62.1, 38.9. LC/MS/MS (ESI) m/z: requires 179.24; found 180.05 ([M + H]⁺, 60), 161.95 ([M–OH]⁺, 100).

2-(3-Bromo-4-isothiocyanatophenyl)ethanol (I2)

The intermediate, 2-(4-amino-3-bromophenyl)ethanol was synthesized according to the literature method using 2-(4-aminophenyl)ethanol and NBS with 80% yield.²⁶¹H NMR (600 MHz, DMSO-d₆, ppm): 7.16 (s, 1H), 6.88 (dd, J = 8.2 and 1.2 Hz, 1H), 6.68 (d, J = 8.2 Hz, 1H), 5.06 (s, 2H, –NH₂), 4.56 (t, J = 5.0 Hz, 1H, –OH), 3.48 (m, 2H), 2.52 (t, J = 7.0 Hz, 2H). ¹³C NMR (150 MHz, DMSO-d₆, ppm): 144.1, 132.5, 129.3, 129.1, 115.7, 107.8, 62.8, 38.0.

Then, compound I2 was obtained from the brominated intermediate and CSCl₂ by following a similar procedure to that described for compound I1. The crude product was purified by column chromatography (silica gel) with hexane : ethyl acetate (1 : 4) to give an orange oily liquid with 60% yield. IR (υ/cm⁻¹): 3334 (O–H stretching), 3063, 3033 (aromatic C–H stretching), 2945–2875 (aliphatic C–H stretching), 2042 (N C S stretching), 520 (C–Br stretching). ¹H NMR (400 MHz, DMSO-d₆, ppm): δ 7.60 (d, J = 1.8 Hz, 1H), 7.43 (d, J = 8.2 Hz, 1H), 7.27 (dd, J = 8.2 and 1.8 Hz, 1H), 4.64 (s, broad, 1H, –OH), 3.57 (t, J = 6.6 Hz, 2H), 2.69 (t, J = 6.6 Hz, 2H). ¹³C NMR (100 MHz, DMSO-d₆, ppm): δ 142.6, 136.0, 133.9, 130.2, 128.1, 127.6, 120.1, 61.8, 38.4. LC/MS/MS (ESI) m/z: requires 258.13; found 257.05 ([M–H]⁻, 44), 241.00 ([M–OH]⁻, 84), 227.15 ([M–OH–CH₂]⁻, 76), 223.00 (100), 201.10 ([M–NCS]⁻, 78).

5-(2-Hydroxyethyl)-2-isothiocyanatobenzonitrile (I3)

The intermediate 2-amino-5-(2-hydroxyethyl)benzonitrile was synthesized according to the literature method using 2-(4-amino-3-bromophenyl)ethanol and nickel(ii)cyanide tetrahydrate in NMP by applying 10 min microwave irradiation (200 °C–200 watt) with 56% yield.²⁶¹H NMR (600 MHz, DMSO-d₆, ppm): 7.18 (s, 1H), 7.14 (d, J = 8.5 Hz, 1H), 6.69 (d, J = 8.5 Hz, 1H), 5.81 (s, 2H, –NH₂), 4.57 (t, J = 5.2 Hz, 1H, –OH), 3.48 (m, 2H), 2.52 (t, J = 6.9 Hz, 2H). ¹³C NMR (150 MHz, DMSO-d₆, ppm): 150.3, 135.4, 132.3, 127.5, 118.8, 115.7, 93.7, 62.4, 37.8.

Then, compound I3 was obtained from the cyano-containing intermediate and CSCl₂ by following a similar procedure to that described for compound I1. The crude product was purified by column chromatography (silica gel) with hexane : ethyl acetate (1 : 3) to give an orange solid with 60% yield. IR (υ/cm⁻¹): 3356 (O–H stretching), 3073, 3037 (aromatic C–H stretching), 2948, 2878 (aliphatic C–H stretching), 2229 (C N stretching), 2022 (N C S stretching). ¹H NMR (400 MHz, DMSO-d₆, ppm): δ 7.77 (d, J = 2.0 Hz, 1H), 7.60 (dd, J = 8.3 and 2.0 Hz, 1H), 7.54 (d, J = 8.3 Hz, 1H), 4.67 (s, broad, 1H, –OH), 3.58 (t, J = 6.5 Hz, 2H), 2.74 (t, J = 6.5 Hz, 2H). ¹³C NMR (100 MHz, DMSO-d₆, ppm): δ 139.3, 138.3, 138.1, 135.1, 134.4, 133.8, 127.5, 117.0, 61.8, 38.2. LC/MS/MS (ESI) m/z: requires 204.25; found 204.00 ([M–H]⁺, 100), 180.05 ([M–CN]⁺, 70), 163.00 ([M–CN–OH]⁺, 96).

General procedure for the synthesis of ester derivatives (I1a–e)

To a solution of compound I1 (1.0 equiv.) in DCM were added pyridine (1.2 equiv.) and acyl chloride derivative (1.2 equiv.) in an ice bath. The mixture was stirred for 15 min and then at room temperature for 8–24 h. The solvent was removed in vacuo. The crude products were purified by flash chromatography on silica gel (hexane/ethyl acetate = 4 : 1 or 5 : 1) to give target compounds with a yield of 50–75%.

4-Isothiocyanatophenethyl benzoate (I1a)

The reaction was performed under general procedure conditions using compound I1 and benzoyl chloride. White solid, m.p. 75.5–77 °C. IR (υ/cm⁻¹): 2113 (N C S stretching), 1711 (ester C O stretching). ¹H NMR (400 MHz, DMSO-d₆, ppm): 7.89 (d, J = 8.2 Hz, 2H), 7.64 (t, J = 7.0 Hz, 1H), 7.51 (t, J = 7.4 Hz, 2H), 7.40 (m, 4H), 4.48 (t, J = 6.3 Hz, 2H), 3.06 (t, J = 6.3 Hz, 2H). ¹³C NMR (100 MHz, DMSO-d₆, ppm): δ 166.1, 139.0, 133.9, 133.6, 130.9, 130.1, 129.6, 129.3, 128.7, 126.5, 65.4, 34.5. HRMS [C₁₆H₁₃NO₂S + Na]⁺ requires 306.0565; found 306.0549.

4-Isothiocyanatophenethyl 4-methoxybenzoate (I1b)

The reaction was performed under general procedure conditions using compound I1 and 4-methoxybenzoyl chloride. White solid, m.p. 81–82.5 °C. IR (υ/cm⁻¹): 2126 (N C S stretching), 1699 (ester C O stretching). ¹H NMR (400 MHz, DMSO-d₆, ppm): 7.81 (d, J = 8.8 Hz, 2H), 7.35 (m, 4H), 6.99 (d, J = 8.8 Hz, 2H), 4.40 (t, J = 6.5 Hz, 2H), 3.79 (s, 3H), 3.01 (t, J = 6.5 Hz, 2H). ¹³C NMR (100 MHz, DMSO-d₆, ppm): δ 165.7, 163.7, 139.1, 133.6, 131.7, 130.9, 128.7, 126.4, 122.3, 114.6, 65.0, 56.0, 34.5. HRMS [C₁₇H₁₅NO₃S + Na]⁺ requires 336.0670; found 336.0655.

4-Isothiocyanatophenethyl 4-fluorobenzoate (I1c)

The reaction was performed under general procedure conditions using compound I1 and 4-fluorobenzoyl chloride. White solid, m.p. 69–70 °C. IR (υ/cm⁻¹): 2117 (N C S stretching), 1713 (ester C O stretching). ¹H NMR (400 MHz, DMSO-d₆, ppm): δ 7.93 (q, J = 8.6 and 3.0 Hz, 2H), 7.36 (m, 4H), 7.31 (t, J = 8.8 Hz, 2H), 4.44 (t, J = 6.5 Hz, 2H), 3.03 (t, J = 6.5 Hz, 2H). ¹³C NMR (100 MHz, DMSO-d₆, ppm): δ 166.9 and 164.4 (d, J = 251.3 Hz), 165.2, 138.9, 133.6, 132.5 and 132.4 (d, J = 9.6 Hz), 130.9, 128.7, 126.75 and 126.73 (d, J = 2.4 Hz), 126.5, 116.6 and 116.4 (d, J = 22.2 Hz), 65.5, 34.4. LC/MS/MS (ESI) m/z: requires 301.34; found 300.15 ([M–H]⁻, 100).

4-Isothiocyanatophenethyl 4-nitrobenzoate (I1d)

The reaction was performed under general procedure conditions using compound I1 and 4-nitrobenzoyl chloride. White solid, m.p. 142–143.5 °C. IR (υ/cm⁻¹): 2129 (N C S stretching), 1721 (ester C O stretching). ¹H NMR (400 MHz, DMSO-d₆, ppm): 8.31 (d, J = 8.5 Hz, 2H), 8.10 (d, J = 8.5 Hz, 2H), 7.38 (m, 4H), 4.52 (t, J = 6.4 Hz, 2H), 3.07 (t, J = 6.4 Hz, 2H). ¹³C NMR (100 MHz, DMSO-d₆, ppm): δ 164.7, 150.8, 138.8, 135.5, 133.7, 131.1, 130.9, 128.8, 126.5, 124.5, 66.2, 34.3. HRMS [C₁₆H₁₂N₂O₄S + Na]⁺ requires 351.0415; found 351.0410.

4-Isothiocyanatophenethyl 3-methylbutanoate (I1e)

The reaction was performed under general procedure conditions using compound I1 and isovaleryl chloride. Colorless liquid, IR (υ/cm⁻¹): 2053 (N C S stretching), 1733 (ester C O stretching). ¹H NMR (400 MHz, DMSO-d₆, ppm): 7.31 (m, 4H), 4.19 (t, J = 6.6 Hz, 2H), 2.87 (t, J = 6.6 Hz, 2H), 2.08 (d, J = 7.2 Hz, 2H), 1.88 (m, 1H), 0.80 (d, J = 6.7 Hz, 6H). ¹³C NMR (100 MHz, DMSO-d₆, ppm): δ 172.6, 138.9, 133.6, 130.8, 128.7, 126.4, 64.2, 43.1, 34.5, 25.7, 22.6. HRMS [C₁₄H₁₇NO₂S + Na]⁺ requires 286.0878; found 286.0863.

Biological evaluation

COX inhibition assay

The COX-1 and COX-2 inhibitory activities of ITC derivatives were evaluated using a COX (human) Inhibitor Screening Assay Kit (Cayman Chemical) as described.³⁸ Briefly, ITC derivatives were pre-incubated with recombinant human COX-1 or COX-2 enzyme at 37 °C for 10 min. At the end of pre-incubation, arachidonic acid (100 mM) was added to each tube and incubated at 37 °C for 30 s. The reaction was stopped by the addition of stannous chloride. The percentage inhibition of compounds was calculated according to the manufacturer's instructions. All compounds were dissolved in DMSO for all the biological studies.

Cell cytotoxicity assays

The cytotoxicity effects of ITC derivatives on RAW 264.7 cells were determined by MTT assay.⁴⁵ Briefly, cells were seeded in a 24-well cell culture plate at 4 × 10⁵ cells per mL; the plate was incubated at 37 °C for 24 h. At the end of incubation, cells were treated with the compounds at 10 and 20 μM doses, then the plate was incubated for 24 h. After incubation, MTT solution (20 μL) was inoculated in each well and incubated for 2 h. At the end of the period, the medium was aspirated, 200 μL DMSO was added into the wells and cells were incubated at 37 °C for 10 min. The survival of the cells was measured based on the absorbance at 570 and 620 nm using a microplate reader (Tecan Infinite® 200 PRO, Switzerland).

H₂S release of compounds

The H₂S-release capacities of the ITC derivatives were tested as described by Li et al. with small modifications.⁴⁶ Briefly, a reaction mixture containing 20 mM N,N-dimethyl-1,4-phenylenediamine sulfate salt (200 μL) and 1% (w/v) Zn(OAc)₂ (100 μL) were reacted with 30 mM FeCl₃ (200 μL). Then, the ITC derivatives (100 μM in final) were diluted in 1 mM l-cysteine containing phosphate buffer (pH = 7.4). The reaction mixture (500 μL) and the mixture of l-cysteine and derivatives (1 mL) were taken into a 24-well plate. The plate was read on a microplate reader (Tecan Infinite® 200 PRO, Switzerland) at 670 nm at 5-min intervals until a meaningful increase was not observed.

Statistical analysis

Data are obtained by at least three independent biological replicates and expressed as ± standard error of the mean. The differences between treatment groups were assessed by one-way ANOVA with Dunnett's post hoc test (*p < 0.02, **p < 0.005, ***p < 0.001).

In silico studies

Preparation of ligands and target proteins for molecular docking

Initially, the ITCs were sketched using the 2D Sketcher of Maestro. In order to use ligands in docking studies, the 2D structures of the compounds were transformed into optimized energetically minimized 3D structures,⁴⁷ using Maestro's LigPrep module (LigPrep, 2009, Schrodinger LLC, New York, NY⁴⁸) utilizing the OPLS3e force field.⁴⁹ On the other hand, the target proteins utilized in our study were prepared as given in our previous study.³⁸ Briefly, X-ray crystal structures (PDB ID 6Y3C) for COX-1 (ref. 50) and (PDB ID 5KIR) for COX-2 (ref. 51) were obtained from the Protein Data Bank. For the COX-2 protein, only the co-crystallized ligand (rofecoxib⁵²) bound to the ‘A’ chain of the protein structure was retained, while all other ligands, solvent molecules, and ions were carefully removed to eliminate any potential interference in the docking studies. This ensured that the focus remained solely on the interactions between the synthesized compounds and the target binding site. The human COX-1 protein used was in its apo form, thereby providing an unoccupied active site for the docking simulations. Subsequently, another COX-1 protein from Ovis aries (PDB ID 3KK6) that contained the co-crystallized ligand (celecoxib) in the binding pocket was aligned with the 6Y3C-coded structure. Following the alignment, the celecoxib ligand from the O. aries COX-1 structure was carefully transferred and merged into the binding pocket of the human COX-1 structure 6Y3C. As a result of this procedure, we successfully prepared two docking-ready protein models: a human COX-1 protein containing the celecoxib ligand and a human COX-2 protein containing the rofecoxib ligand. These prepared structures served as the basis for the subsequent docking studies, with the binding sites of the crystallized ligands being accurately identified and set as the centers of the grid boxes used in the simulations. The default Glide grid generation settings were employed, utilizing a cubic inner box with equal side lengths. Consequently, the grid boxes for the COX-1 protein were configured with inner dimensions of 15 Å × 15 Å × 15 Å per side and outer dimensions of 28.179 Å per side, while for the COX-2 protein, the grid boxes were configured with inner dimensions of 10 Å × 10 Å × 10 Å Å per side and outer dimensions of 22.786 Å per side. The preparation of proteins and the grids was carried out using Schrödinger's Maestro molecular modeling package. The OPLS3e force field was employed for restrained minimization with a heavy atom convergence of 0.3 Å. Disulfide bonds were formed, and any missing side chains were rectified using the Prime module.⁵³ Protonation states were calculated using PROPKA⁵⁴ at a physiological pH of 7.4, ensuring that the ionization states of the amino acids were appropriate for a biological environment. To ensure completeness, any missing hydrogen atoms and side chains were inserted as needed. After these adjustments, the protein systems underwent thorough energy minimization utilizing the OPLS3 force field. This process was essential to refine the protein structures, remove any steric clashes, and optimize them for subsequent docking studies. The crystallized ligand binding sites on the target proteins were then determined as grid boxes, with the removal of any ions or small elements introduced for crystallization purposes. Upon the successful completion of these preparations, the docking studies were conducted in accordance with our previous study.³⁸ Briefly, the default settings of Maestro Glide were utilized for the docking studies, with the additional option to account for partial charges. Within these settings, the following procedure was implemented: initially, up to 5000 poses were generated for each ligand using a scoring window of 100. From these, the top 400 poses were selected for energy minimization. Post-docking minimization was conducted, with a maximum of 100 energy minimizations allowed after which 5 poses per ligand were retained. Ultimately, only the output pose with the best docking score was selected for each ligand.

Molecular docking studies and MD simulations

The ITCs that were carefully prepared were subsequently subjected to molecular docking with the human COX-1 and COX-2 target proteins. To accomplish this, we employed the grid-based Glide docking program,⁵⁵ with standard accuracy settings. In order to mimic the physiological environment, an explicit water model (i.e., TIP3P) was used. The neutralization of the systems was achieved by introducing 0.15 M NaCl solution. Throughout the simulations, a constant temperature of 310 K and a pressure of 1.01325 bar were maintained using the Nose–Hoover thermostat⁵⁶ and Martyna–Tobias–Klein⁵⁷ protocols, respectively. To investigate the dynamic behavior and stability of the complexes, MD simulations were conducted using Maestro's Desmond module.⁵⁸ Each complex underwent an explicit all-atom simulation study spanning 100 ns, enabling a comprehensive analysis of their behavior and interactions. The complexes with the highest scores were subsequently subjected to further examination regarding their free binding energy through MM/GBSA analysis using Prime software.⁵³

ADME analysis and binary-QSAR models

To evaluate the ADME properties of the synthesized compounds, the QikProp module within the Maestro suite was employed, which provides predictions of ADME parameters. For the assessment of therapeutic activity, the MetaCore/MetaDrug platform (https://portal.genego.com/) developed by Clarivate Analytics was utilized. The MetaCore/MetaDrug platform, specifically designed for therapeutic activity predictions, utilizes disease-QSAR models to evaluate synthesized compounds. Synthesized compounds were evaluated in terms of therapeutic activity predictions using binary QSAR models. Here, we used cancer-QSAR and inflammation-QSAR models. The used model's performance was confirmed by diverse statistical parameters such as sensitivity, specificity, accuracy, and the Matthews correlation coefficient (MCC), allowing for an accurate assessment of their quality. Cancer-QSAR model: training set N = 886, test set N = 167, sensitivity = 0.89, specificity = 0.83, accuracy = 0.86, and MCC = 0.72; inflammation-QSAR model: training set N = 598, test set N = 93, sensitivity = 0.86, specificity = 0.84, accuracy = 0.85, MCC = 0.69. The predictive therapeutic efficacy is determined by ChemTree's capacity to link structural descriptors to said property through the recursive-partitioning algorithm. Optimal results were achieved with the following ChemTree parameters: path length 5, max segments 3, p-value threshold Bonferroni 0.99, p-value multiway split 0.99 and number of random trees 50. In the MetaCore/MetaDrug, predicted therapeutic activity values of the models are normalized between 0 and 1 (i.e., 0, inactive; 1, active).

Data availability

The data that support the findings of this study are available in the ESI† of this article.

Author contributions

Design and supervision of the research: T. B. T., T. G. and M. A. Synthesis/characterization of the compounds and analyses of data: T. G., Y. B. Y. and M. A. Bioactivity studies: Y. B. Y, Serhat D., H. N. A and T. B. T. In silico studies: S. D. and P. S. Writing – original draft preparation: all authors. Review and editing: T. B. T., T. G., M. A. and S. D.

Conflicts of interest

The authors declare no conflicts of interest.